Skip to main content

Full text of "Case and Agreement in Abaza"

See other formats

1.1.2 Morphology and syntax 

Abaza is highly agglutinative, involving a great deal of morphological 
alternation, particularly in the verb complex and, to a lesser degree, in the 
noun complex. S3nitactically, Abaza is strongly head-final. The following 
sections discuss the basic morphological alternations and the basic syn- 
tactic structure of postpositional phrases (, nominal phrases 
(, and verbal phrases ( Paradigms for the agreement of a 
postposition with its object, a noun with its possessor, and a verb with its 
subject, object, and indirect object are not presented here, but can be 
found in chapter 2. Postpositional phrases. As the name suggests, postpositions 
follow their complements- Agreement with the object of the postposition 
is registered with a prefix on the postposition. Postpositions are mostly 
locative and temporal, but there are comitative and benefactive postposi- 
tions, as well. Many relations expressed by prepositions or postpositions 
in other languages are expressed by prefixes in the verb complex. (See 
chapter 7 for a treatment of these verbal prefixes as incorporated 

Example (2) shows a temporal postposition with a third-person singular 
irrational or inanimate complement, i.e., object. Example (3) shows a 
locative postposition with a first-person singular complement. 

(2) aw2y a-mftax^ 
that 3si-after 
after that 

(3) sara s-prid 
I Is-at 

at my house, by me Nominal phrases. Abaza nouns morphologically mark a vari- 
ety of grammatical distinctions, including number, deflniteness, and 
agreement with a possessor. Adjectives are generally compounded with 
the noun. Conjimction is marked morphologically on each conjunct. 

•^spruit (1986) offers a detailed account of stress in Abkhaz, which is similar to Abaza. 

1 Introduction and Background 

Plurality is marked with the suffix -fc^a, as in (4) and (5). In addition, a 
second plural suffix, -c^a, may be used for rational (human) nouns, either 
alone or between the root and the suffix -fc^a, as in (6). For some words re- 
ferring to the young of animals, and ending in [^s] or [is] {/9ys/), this end- 
ing is replaced by [ara] to form the plural, as in (7). Except for cases like 
(7), there is no overt marking to indicate nonplurality. 

(4) a-sab9y-k'^a 
the children 

(5) ha-fc**'a 

(6) y-a/^c^a-fc*^a 
his brothers 

(7) a. it^c'^-as 


b. k\'-ara 

The kind of information carried by determiners in other languages is 
signaled morphologically on the noun in Abaza. The definite marker is the 
prefix a-. It appears to be used in a wider range of contexts than the defi- 
nite article in English. The indefinite marker is the suffix -fc'. 

(8) a. a-bmhard^"^ 

the butterfly 

b. bm/iaraf^-fc' 
a butterfly 

1 , 1 Grammatical sketch of Abaza 1 1 

Deictics and demonstratives, such as away 'this' and arsy *that', are inde- 
pendent words md occur to the left of the head noun, as in (9).^^ Quantifiers, 
such as Z9mr^a *ill, every' and mac'^ *few', occur to the right of the head noun, 
as in (10). In this way they behave as adjectives. These determiners can also 
all be used prorominally. 

(9) ardy a-qac'a 
that DEF-man 
that man 

^10) a. g'^sp Z9m^a^'* 
group all 
the whole group 

b. a-qac'a-k^a zdm^'^a-ta 
DEF-man^PL all-AV 
all the men 

The plural and indefinite suffixes may co-occur, with the interp^-etation 
'some'. The indefinite suffix occurs outside the plural suffix. 


some things 

The suffix -k' is also used to indicate the number *one', sO that (8b) can 
also mean *one butterfly'. Numerals between two and ten are indicated by 
prefixes, plus the obligatory suffix -k\ This suffix differs from the indefi- 
nite suffix in its distributional properties in that it may occur with the def- 
inite prefix and within stative predicates. 

(12) ^''-q'arTm-k' 

two daggers 

'^The definite prefix is identical to the third-person singular irrational (nonhiunan or 
inanimate) agreement marker. Thus, the form in (9) could also mean 'that (irrational) one's 
man'. It is likely that the definite marker either developed from the agreement marker, or 
that a definite reading is one use of this marker. 

''*In the dialect I am most familiar with, which is unwritten, zamf^a tends to form a 
phonological unit with the head noun. The literary dialect in the former Soviet Union 
(T'ap'anta) writes zam5^a as a separate word, though adjectives may be written together 
with their respective nouns (as compounds). In either case, the quantifiers occupy a 
position to the immediate right of the noun. 

1 2 Introduction and Background 

(13) X'q'ama-k' three daggers 
pf~q 'ama-k ' four daggers 
jc**'-q 'ama-k ' five daggers 
c-q'ama~k* six daggers^ ^ 
b93^-q'ama-k' seven daggers 
^a-q 'ama-k ' eight daggers 
S'^-q'ama-k' nine daggers 
S'^a-q'ama-k* ten daggers 

Possessors occur to the left of the head noun, and agreement with the 
possessor is registered on the head noun with a prefix in the ergative 
agreement series. Examples of possessed NPs can be seen in (14) through 

(14) ahmet y-tdzd 
Ahmet 3sm-house 
Ahmet's house 

(15) ac'ays a-?^ara 

DEF-bird 3si-nest H 

the bird's nest 

(16) a-phas lqac*a 
DEF-woman 3sf-man 
the woman's husband 

The definite article may not co-occur with either the numerals or the 
possessive prefixes, but the possessive prefixes may co-occur with the nu- 
merals. In that case, the possessive prefix occurs outside, i.e., to the left of, 
the numeral. 

(17) sosr9q*^a yd-^^-famqa-k' 
Sosruko 3sm-two-knee-iNDEF 

Sosruko's two knees nart, 51^^ 

Modifiers (adjectives) are suffixed to the noun root, forming a com- 
pound. Plural and indefinite morphology occurs to the right of adjectives. 
The adjective is compounded with its respective head in (18) through 
(20), demonstrating its position inside both the indefinite marker and the 

'^The prefix for *six' is a rapid diagnostic for differentiating between Abaza and Abkhaz. 
It is c in Abaza, and /in Abkhaz. 

'^See abbreviations list for source of cited examples. 

1 . ? JrwTvnatical sketch of Abaza 1 3 

plural marker. Multiple adjectives may also be compounded to the right 
of the noun head, as in (21) and (22). 

(18) C9-(bw-k' 
a big horse 

(19) a-mj^-daw-k^a 
the big bears 

(20) qac'a-bz9y'k'^a 
friendly men 

(21) q'^marga-pjdza-ibw 
pretty big toy 

(22) wasa-k'^ayc'^a'psdla-dBW 
big plump black sheep 

A second way of forming nominal modification is as an adverchial ad- 
junct, marked with the adverbializer ta, as in (23). This construction rlas a 
more limited distribution. Note that in both types of adjectival modifica- 
tion, the modifier is to the right of the noun head, which is unexpected 
considering the strong preference for head-final structures in the 

(23) ca-fc' d9W-ta 
horse-iNDEF big-AV 
a big horse 

Relative clauses precede their heads, as in (24). When both possessor 
and relative clause are present, the possessor precedes the relative clause, 
as in (25). 

(24) x-wasa-k' z-p9-z a-la 
three-sheep-lNDEF EWH-kill-PST DEF-dog 
the dog that killed three sheep 

14 Introduction and Background 

(25) ahmet yd-m-ca-wa y9-wand9r 
Ahmet AWH-NEG-go-PTC 3sm-car 
Ahmet's car that doesn't work 

In addition to these affixes, conjunction is marked by the suffix -y on 
each conjunct. This is also true if there are more than two conjuncts. 

(26) s-am-y s-aha-y 

1 s-mother-and 1 s-father-and 
my mother and father 

(27) s-ana-y s-aha-y s-apa-y 

1 s-mother-and 1 s-father-and 1 s-brother-and 
my mother and father and brother 

The conjunction -y can be supplemented with the intensive suffix -g^d, 
which likewise occurs on each conjunct. 

(28) s~an9-g^dy s-aba-g^B-y 
Is-mother-lNT-and Is-father-lNT-and 
my mother and father 

(29) s-ana-g^d-y s-aba-g^d-y s-afa-g^d-y 

1 s-mother-lNT-and 1 s-father-iNT-and 1 s-brother-iNT-and 
my mother and father and brother 

It is possible for the intensive suffix -g^9 to occur on only the final con- 
junct, but it has a slightly different interpretation, in which the final con- 
junct is somehow separate from the other(s). 

(30) s-and-y s-aba-y s-afa-^d-y 

1 s-mother-and 1 s-father-and 1 s-brother-INT-and 
my mother and father, and my brother (separately) 

These conjunctions are the only nominal affixes shown here which can 
also occur with pronouns. 

(31) sara-y wara-y 
Is-and 2sm-and 
you and I 

1 . 1 Grammatical sketch of Abaza 1 5 

'^ Verbal phrases. The most basic distinction in the Abaza ver- 
iDal system is between dynamic and stative verbs. These have independent 
systems of affixation. Note that predicates of nonverbal categories take 
morphology from the stative system. See chapter 3 for a discussion of 
stadve predicates. 

Dynamic verbs. The dynamic verb in Abaza may host a wide variety of af- 
fixes. Prefixes include various agreement markers, preverbs, directionals, 
causative, potential, and markers of various oblique relationships. Suffixes 
include tense, aspect, and mood markers. Negation occurs as either a prefix 
or a suffix. 

Prefixes, One of the most salient sets of prefixes in the Abaza verbal sys- 
tem is the set of agreement markers. The verb in Abaza registers agree- 
ment with all of its arguments, including subject, object, and indirect 
object. Agreement morphemes are all prefixes, with the series that regis- 
ters agreement with intransitive subjects and transitive direct objects 
(absolutive) occurring to the left of the other basic agreement series 
(ergative), which is used to register agreement with transitive subjects, 
indirect objects, causative subjects, (see chapter 4), and various oblique 
relaionships including benefactive, instrumental, comitative, and vari- 
ourlocatives (see chapter 7). (See chapter 2 for a discussion and analysis 
of the basic agreement system.) 

Reflexive and reciprocal constructions are signaled morphologically by 
P'efixes. Reflexivity is indicated by the use of the prefix c-, which occurs 
in the same position as the absolutive series of prefixes. The reflexive pre- 
f X replaces the absolutive prefix which would otherwise agree with the 
object. The subject continues to be registered in the ergative series. (See 
sections 4.2 and 6.4 for further discussion of reflexives.) 

(32) c-s-fc'*aba-d 

I bathed myself. 

Reciprocity is indicated by one of several reciprocal prefixes, the most 
common being aba-, (The choice of reciprocal prefix is lexically deter- 
mined.) The subject is registered in the absolutive series, and there is 
nothing other than the reciprocal prefix to indicate the presence of a di- 
rect object. In this sense, the reciprocal prefix detransitivizes a transitive 
verb, while a reflexive does not. For presumably pragmatic reasons, the 
reciprocal occurs only with plural subjects. 

1 6 iTttroduction and Background 

(33) h-aba-ba-d 
We saw each other. 

(34) f-aba-ba-d 

You (PL) saw each other. 

(35) y-abaM-d 
They saw each other. 

Some verbs have complex roots. For these verbs with two distinct parts, 
various elements, including the ergative agreement marker, may occur 
between the two parts, but both are obligatory. This can be seen, for ex- 
ample, with the complex root pq *break\ Following the tradition of 
Caucasianists, I refer to the part of the verb to the left of the ergative 
marker as the preverb, and the part to the right as the verb (root). In gloss- 
ing examples, I simplify by glossing preverbs as PV, and assigning the 
overall meaning of the preverb-root combination to the root. In actuality, 
it is the combination of preverb plus root which gives the overall mean- 
ing. In examples in which the preverb and root are contiguous, I generally 
do not separate them into preverb and root, but give a single gloss for the 

(36) y-p-s-QB-d 
I broke it. 

The preverbs have no independent (compositional) meaning. Some 
roots, or SUFFIXOIDS (Klych 1988), however, may occur with different 
preverbs, which results in different complex verbs. Certain roots may also 
occur without a preverb, but with a distinct meaning from the counterpart 
having a preverb. (For more discussion of preverbs, see K'lych 1988 and 
Klychev 1994.) 

Other prefixes occurring in the verbal complex include the directional 
markers, fa- and na-. These prefixes indicate action towards and away 
from the speaker, respectively, although the reference point of the 
directionality is not absolutely fixed. They occur between the ergative 
and absolutive agreement series. These prefixes are common, especially 

1 . 1 Grammatical sketc^^ ofAbaza 1 7 

fa-, and highly productive. They also serve various discourse functions. 
(See O'Herin 1985 for a discussion of the discourse functions.) 

(37) a. fagfra ( < Sa-ga-ra DiR-carry-MAS) to bring 
b. imgra ( < na-ga-ra DiR-carry-MAS) to take 

There is a set of spatial prefixes with more specific locational meanings 
than the directionals. Examples, taken from Genko (1955:171-172), are 
given in (38). None of these are ever associated with agreement or an 
overt argument. 

(38) Jt'a- 

movement downwards 

)**'ak^-, )**'aZ- 

movement outwards 

ck'ara- (ck'ar?-) 

in, from the corner 

pt'a- (pt9-) 

down flat 

pqa- (pq9-) 

forward, iii front of 

era- (cn~) 

at, from the hearth 

ta- (tB-) 

into, inside 

bja- (bj9') 

amidst, from among 


on horseback, by horse 


on the shore 


before, in front of ^^ 


behind (location) 


at home, inside 

fc'ala- (k'9l) 



behind (direction) 

Example (39) shows the prefix k'a- in a fully inflected form. 

(39) a-cd awa?a y-k'a-ha-d 

DEF-horse there 3si-down-fall-DYN 

The horse fell there. NART, 89 

Preverbs, directionals, and these spatial prefixes share the property that 
there is never agreement associated with any of them. They occur in the 
same general morphological space between the absolutive and ergative 
agreement series. Although there are minor differences in position, there 
seems to be a relatively fixed order among them. The primary difference 
between the preverbs, on the one hand, and directionals and spatial 

•^Some Abaza speakers have glossed this as ^against*, so that it contrasts with pqa-. See 
pVa-g^l-ra 'resist (Ut., against-stand-MAS)\ 

18 Introduction and Background 

prefixes on the other, is that the preverbs are obhgatory, while the others 
are optional. The directionals differ from the spatial prefixes in that they 
are much freer in their distribution and in that they may serve certain dis- 
course functions. In a few cases, directionals or spatial prefixes occur as 
obligatory preverbs.^^ 

Certain oblique functions can also be expressed in the verb complex, in- 
cluding the benefactive, the comitative, the instrumental, and various 
locatives. These are indicated through the use of prefixes and, as already 
mentioned, there is agreement with the oblique argument. (See chapter 7 
for a treatment of these as incorporated postpositions.) 

(40) y-s-z-a-c^a-d 

It did it for me. SSAVC, 163 

The causative prefix, r-, increases the number of arguments registered 
through agreement by one. The causative subject is registered in the 
ergative series to the immediate left of the causative morpheme. (See 
chapter 4 for a discussion of the causative construction.) 

(41) d-q'^BC-i-t' 

S/he thinks. GAL, 178 

(42) d-a-r-q'^dc-i-f 

It made him/her think. GAL, 178 

The potential prefix, z-, indicates permission or ability for an action to 
be performed. (For a discussion of the potential construction, see chapter 

(43) yd-s-h'^a-y-t' 
I say it. 


3si-l s-POT-say-PRS-DYN 

I can say it. GAL, 116 

**I assume that these are in transition between the two relevant types. 

i. 1 Grammatical sketch ofAbaza 1 9 

Schematically, the dynamic verbal prefixes can be seen in table 2. The 
absolutive and ergative agreement series (ABS, ERG) are interspersed with 
the other categories. The ergatives and the obliques (OBL) may include 
multiple prefixes of the same type. The column indicated by LOG includes 
both locative obliques and the spatial prefixes. The obliques and locative 
obliques involve agreement with the oblique argument in addition to the 
oblique prefix itself. Sample morphemes are given under each type. 

Table 2. Order of dynamic verbal prefixes 


S-, b-, Z-, ^a-, p-, ta-, k'a-; s-, b-, z- s-, b-, r- V 
W-, J"*-, na- ... k"*-, W-, 1-, w-, 1-, 

d-,... la-, c- dzqa-, 

Null absoliidve agreement In one limited set of cases, absolutive agree- 
ment is obligatorily missing where one would otherwise expect it to be 
present in the verb complex. Additionally, in these cases the nominal and 
the verb complex tend to form a tighter phonological (prosodic) unit than 
when the absolutive marker is present, so that [a^dzas'^d] forms a single 
stress unit in (45).^^ 

(45) a-la a-dz a-s'^-d 
DEF-dog DEF-water 3si-drink-DYN 
The dog drank the water. 

According to Allen (1956:134), Tabulova (1976:113), and Sergei Pazov 
(pers. comm.), both of the conditions in (46) must be met for absolutive 
agreement to be absent. 

(46) a. the nominal with which the missing absolutive marker would 

register agreement is either 3si or 3p (the two absolutive 
forms which have the phonological form y),^° and 
b. the nominal with which the missing absolutive marker would 
register agreement must occur immediately before the verb 

'''See Allen (1956) for a fuller discussion of the phonological properties of this 

20The + wh form, which is also phonetically ya, does not participate in this alternation. It 
is further distinguished from the other two y forms phonologically in its stress properties. 
See O'Herin (1992b, 1994) for further discussion of stress in Abaza. 

20 Introduction and Background 

This contrast can be seen in (47M50). In (47), the conditions are met 
for null absolutive agreement, and the absolutive agreement marker is 
null. In (48), the conditions are met for null absolutive agreement, but 
there is an overt absolutive prefix. Such forms are highly disfavored, but 
marginally acceptable, especially in very slow speech. In (49), the condi- 
tions are not met for null absolutive agreement because the adverb /a/ta 
'early' intervenes between the direct object and the verb, so the 
absolutive prefix is obligatory. Contrast (50), in which the conditions for 
null absolutive agreement are not met, but in which the absolutive prefix 
is missing. This leads to full ungrammaticality. 

(47) sara a-mj^ s~ba-y-t* 

I DEF-bear Is-see-PRS-DYN 

I see the bear. 

(48) ??sara a-mj^ ya-5-6a-y-t' 

I DEF-bear Sgi-ls-see-PRS-DYN 

I see the bear. 

(49) sara a-nf^ fafta y^-s-ba-y-t' 

I DEF-bear early 3si-ls-see-PRS-DYN 

I see the bear early. 

(50) *sara a-mj^ fafta s-ba-y-t* 

I DEF-bear early 1 s-see-PRS-DYN 

(I see the bear early.) 

It is irrelevant to the generalization in (46) which syntactic position is reg- 
istered by the absolutive prefix. The above examples demonstrate direct ob- 
jects of transitive verbs. Other positions which are registered by absolutive 
agreement include the subject of an intransitive verb, (51) and (52), the di- 
rect object of a ditransitive verb, (53), the subject of an inverted verb, (54), 
and the subject of a stative predication, (55). (See chapter 3 for further dis- 
cussion of stative predicates, and chapter 5 for inverted verbs.) Each of these 
positions allows null absolutive agreement when the corresponding nominal 
occurs to the immediate left of the head bearing this (null) absolutive agree- 
ment, as shown in the following examples. 

(51 ) a-qac 'a-k^a nxa-y-d 
DEF-man-PL work-PRS-DYN 

The men are working. SSAVC, 152 

1.1 Granmmtical sketch ofAbaza 21 

(52) a-s"^ g'^dmxa a-hBmac a-za-faruca-t' 
DEF-cow bad D£F-shed 3si-BEN-remain-DYN 

The shed is the bad cow's lot. GAL, 112 

In (52), the subject of the intransitive verb fanxa 'remain' is ahamai 
'shed' which occurs to the immediate left of the verb. The DP aj**' g^9mxa 
'the bad cow' is a benefactive object. 

(53) $ara wara a-V^a na-w9-s-t9-y-d 

I you DEF-apple DlR-2sm-ls-give-PRS-DYN 

I give you the apple. 

(54) a-fa ^a-s9-cha-t' 

DEF-dog DIR-ls-bite-DYN 

The dog bit me. 

(55) sara t3dzd-k' s-9ma-b 

I house-lNDEF 1 s-have-STP 
I have a house. 

The conditions for null absolutive agreement in (46) are relaxed some- 
what in two circumstances. The overt form, y, may occur in slow speech 
even when the nominal with which it registers agreement occurs to the 
immediate left. This is not unexpected if the rule is dependent (partially 
or wholly) on phonological environment, since an extra pause may be in- 
volved in slow speech that is not present in normal rates. If the pause 
counts as something intervening between the two elements for the pur- 
pose of (46), this conforms to the rule. 

The second situation in which restrictions on (46) are relaxed is in the 
case of stative predicates. (See chapter 3 for further discussion.) Null 
absolutive agreement is often possible even when the nominal with which 
it registers agreement is either null itself or not immediately adjacent to 
the head bearing absolutive agreement. The most common such occur- 
rences are with certain meteorological statements, as in (56) and (57). It 
is also possible, but not required, if the stative predicate begins in a conso- 
nant, especially an obstruent, as in (58) and (59). 

(56) razd-b 

It is foggy. 

22 Introduction and Background 

(57) pxJa?b 
It is hot. 

(58) qYaV~b 

It is the country, homeland. nart, so 

(59) sana-b b-xhz 
sana-STP 3sf-name 

Her name is Sana.^^ NART, 50 

The generalization in (46) is purely dependent on the surface string of 
constituents. The rule must have access to the morphological structure 
(and shape) of the verb complex on the right (to see that the absolutive 
agreement prefix is 3si/3p) and the syntactic structure of the nominal to 
its left (to see that the nominal is that which is registered by the 
absolutive agreement prefix in question). It must be a relatively late rule, 
which also has access to phonological information (to guarantee that the 
two parts form a prosodic constituent). I propose, then, that this rule de- 
letes the 3si/3p absolutive agreement prefix in (or on the way to) the PF 
component under the right conditions. Note that this requires that this 
level of representation have access to syntactic and morphological struc- 
tural information. 

Suffixes. The categories of tense, aspect, and mood are expressed in the 
verbal suffix system.^^ Dynamic verbs distinguish six basic tenses, plus 
various complex tenses. For indicative verbs, the basic tense forms are as 
in (60).^^ 

(60) a. Past completive s-ca-0-d 


I went, 
b. Past incompletive s~ca-w-n 


I was going. 

2 'The subject, hxhz, occupies the anti-topic position to the immediate right of the verb. 
22See Chkadua (1970) for a fuller discussion of tense and mood in Abaza. 
^H will henceforth not indicate null aorist (or other null) affixes. I also will refer to futi 
as just FUT, as this is the considerably more common future tense. 

1, 1 Grammatical sketch ofAbaza 


c. Past dependent 

d. Present 

e. Future definite 

f. Future indefinite 



I was going (and..,) 


I go, am going. 


I will (definitely) go. 


I will go. 

These same tense distinctions hold in other moods, but some forms dif- 
fer slightly in their phonological form, for example the present tense is 
generally -wa in nonindicative moods. 

(61) a. Past completive 

b. Past incompletive 

c. Past dependent 


Did I go? 



Was I going? 


d. Present 

e. Future definite 



Do I go?, Am I going? 


Will I (definitely) go? 

^'fThere is no past dependent yes-no interrogative question form. 

24 Introduction and Background 

f. Future indefinite s<a-waf-ma 

Will I go? 

There are various aspectual suffixes which occur either between the 
root and the tense marker, or between the tense and mood markers, de- 
pending on the individual suffix. These include -x ^repetitive', -rk'^^a 
'continuative', -la ^frequentative', etc. 

(62) a. d-q'^mar-x9'y't' 

S/he plays again. 

b. y-a-z-q'^dCd-rk'^a-y-t' 
They still think about it. 

c. d-q'^ar-la-y-t' 
S/he often plays. 

d. d-ata-ca-jca-y-t' 


S/he goes one more time.^^ 

The possible moods in Abaza include at least indicative, yes-no inter- 
rogative, content-interrogative (rational and irrational), imperative, con- 
ditional, nonfinite, relative, gerund, subjunctive, and desiderative. These 
various moods are demonstrated in (63) all in the aorist tense, which has 
a null tense marker (not marked), where applicable for the tensed forms. 

(63) a. indicative d-ca-d 

S/he went. 

b. yes/no interrogative d-ca-ma 

Did s/he go? 

^^Tnhe morphemes ata- and -X9 form a discontinuous constituent meaning one more time. 

lA Grammatical sketch ofAbaza 


content interrogative 

content interrogative 

,e. imperative 

f. conditional 

g. nortfinite 

h. relative 

i. gerund 

j. subjunctive 

k. desiderative 



Who went? 


What went? 

Go! (PL) 

If sAe goes, ... 



for him/her to go 



(the one) who went 

going, s/he... 

S/he would go. 

Would that s/he go! 

The two types of content questions and the relative form all require ex- 
actly one argument to be registered with wh-agreement. (See chapter 8 
for a fuller discussion.) The relative and gerund forms are identical except 
that the gerund has no [ + wh] arguments. These two forms lack overt 
mood suffixes, but do exhibit normal tense distinctions. 

26 Introduction and Background 

The imperative lacks both tense or mood suffixes. The subject must be 
second person. Of all the moods and tenses, only the imperative causes 
any change in the pattern of agreement markers. If the imperative is posi- 
tive, the subject is singular, and the verb is transitive, the ergative marker 
is obligatorily absent. Otherwise, agreement is as in other moods. This 
obliterates the masculine versus feminine distinction for the imperative of 
transitive verbs. 

(64) W'Ca 
Go! (m) 

(65) b-ca 
Go! (0 

(66) f-ca 
Go! (PL) 



Get (carry) it! (SG) 

(68) y-f-ga 

Get (carry) it! (PL) 

None of the mood suffixes may occur when there is a separate overt 
complementizer, such as asqan 'while, when', as example (69) demon- 
strates. The verb complex anpcd, which is in the clause having the overt 
head asqan, has no mood marker on it. 

(69) a-haq'*' an-pc9 asqan sosrdq^^a awa?a d-^atdj^t-d 
DEF-stone when-break when Sosruko there 3sr-fell-DYN 
When the stone broke, Sosruko fell there. nart, 54 

Schematically, the order of dynamic verbal suffixes can be charted as in 
table 3. 

1.1 Gfwnmatical sketch ofAbaza 27 

Table 3. Order of dynamic verbal suffixes 







X9, rk"^a, 
la, ... 


ra, ... 

Z, ... 

d, b, n, 
ma, ya, 
da, ... 

Masdars, The masdar, a type of nominalization, is used as the citation 
form of verbs.^^ The masdar-forming morpheme, -ra, is a suffix. This suffix 
may also be used with roots of nonverbal categories to give an abstract 
noun, as in (71). 

(70) ca-ra 


to go, going 

(71) d3W-ra 

There are actually two distinct masdar constructions, verbal and nomi- 
nal.^^ Both are formed by the addition of the masdar suffix to a verb. iTiey 
differ in that the nominal masdar may take only nominal modification 
(adjectives to the right of the masdar), while the verbal masdar may take 
only verbal modification (adverbials to the left of the masdar). Compare 
the nominal masdar in (72) with the verbal masdar in (73). 

(72) apx^a-ra-hzdy 
good studying 

(73) larba^'^sga-la ds-s-ba-ra 
glasses-iNST 3sr-ls-see-MAS 
. . . me to see him with glasses 

26 "Masdar" is the traditional Caucasianist term for the derived forms described in this 

2^The verbal masdar is not common. In a text of roughly 6,600 words, there is one clear 
case of a word which is possibly a verbal masdar. (The other option is a future2 participle, 
which has the same phonological form.) There are over a hundred nominal masdars in the 
same text. Most of my examples of verbal masdars are elicited, and these are likewise rare. 

28 Introduction and Background 

They differ also in their agreement patterns. The nominal masdar may 
have only a single ergative agreement prefix, which registers agreement 
with the possessor. The nominal masdar is the only form in the language 
involving a verb root that does not allow absolutive agreement. Compare 
the nominal masdar in (74) with the verbal masdar in (75). 

(74) s-ba-ra 

my seeing, seeing me 

(75) s3-y-ba-ra 
... him seeing me 

For nominal masdars, this ergative agreement series occurs external to 
all other prefixes, regardless of its normal relative order in the verbal pre- 
fix system. The verbal masdar exhibits an agreement pattern identical to 
that in a normal (nonmasdar) verb. Compare the nominal (76) and the 
verbal (77) with respect to the reflexive prefix. 

(76) s-c-k'^aba-ra 

1 S-REFL- wash-MAS 

my self-washing 
i77) c-s-k'^^aba-ra 

REFL- 1 S- wash-MAS 

washing myself, to wash myself 

Finally, the nominal masdar may have nominal morphology, such as 
the plural suffix and definite and indefinite articles. 

(78) a-ca-ra 

the going 

(79) ca-ra-k' 


a going 

i. J Grammatical sketch ofAbaza 


(80) ca-ra-fc"'a 


NegatioTt Negation, which is also morphologically marked in the verb 
complex, occurs in four major patterns, depending on the tense and mood 
of the vert), the four patterns are given in (81). Examples of these four 
patterns are shown in (82). There appear to be two suffix positions for the 
negative morpheme, immediately following the verb root, as in (82a), 
and final, as in (82c).^® 

(81) a. the negative suffix, -m, only, 

b. the negative prefix, m-, only, 

c. the negative suffix, -m, plus the intensive prefix ^-, and 

d. the negative prefix, m-, plus the intensive prefix g'-. 

(82) a. conditional 



if s/he hadn't gone 

b. imperative 

Don't go! 


c. present indicative 

d. aorist yes-no question 



I'm not going. 


Didn't s/he go? 

Nothing more will be said in this study about negation and the intensive 
prefix, as the distribution and analysis is quite complex and not directly 
relevant to the general issues discussed. (See O'Herin 1992a for further 

28When the negative suffix occurs in final position, there is no overt mood suffix. 
29The negative imperative idiosyncratically requires the mood suffix -n in Tapanta. In 
the Anatolian dialects, this has been replaced by a second instance of -m. 

30 Introduction and Back^vund 

Constituent order. Abaza clauses are verb-final. Thus, for intransitive 
verbs, the order is subject-verb. Subjects precede other arguments, so 
transitive clauses are subject-object-verb. In ditransitive clauses, the basic 
order places the indirect object before the direct object, giving the order 
subject-indirect object-direct object-verb. 

(83) ada acta-y-d 

frog croak-PRS-DYN 
The frog is croaking. 

(84) a-c^*'^9n a-h^ah^-k^a y-F^a-y-d 
DEF-boy DEF-dove-PL 3sm-raise-PRs-DYN 
The boy raises doves. 

(85) aw9y sara axc% ^a-s9-y-t-d 

s/he I money DiR-ls-Ssm-give-DYN 

He gave me money. 

There are at least two privileged nonargument positions, including a fo- 
cus position to the immediate left of the verb, as in (86), and an antitopic 
position to the immediate right of the verb, as in (87).^^ The antitopic po- 
sition is apparently reserved for an XP which is already salient in a dis- 
course. In the text containing (87), Sosruko is the main character, and he 
is mentioned by name in the immediately preceding sentence. 

(86) s-k^tab dazda y-na-z-ax"^ 

1 s-book who 3si-DiR-EWH-take 
Who took my book? 

(87) totraf-dyft2-y, pfay-y, wag'^awajh-y d-g^-rd-m-g'^apxa-d 
Totrash-even-and Pshey-and Wagwoj-and Ssr-INT-Sp-NEG-like-DYN 

But Totrash, Pshey, and Wagwoj weren't pleased with Sosruko. 

NART, 57 

Adverbs can occur relatively freely with respect to the arguments. They 
generally occur to the left of the verb complex, but there may also be one 
adverb to the right of the verb complex, possibly in the antitopic position. 

3t*Compare similar structures in Turkish (Hankamer, pers. comm.) and other languages 
(Vallduvi 1990). I treat this as right-adjunction to CP. 

J, 1 Grammatical sketch ofAbaza 31 

Abaza is a pro-drop language, in that pronouns are allowed to be null 
(pro) when their features are indexed on a governing element. Verbal 
agreement cannot be interpreted as pronominal incorporation, since ver- 
bal agreement with arguments is obligatory, regardless of whether there 
is an explicit nominal or not. In addition, Abaza has independent overt 
pronouns, and agreement occurs even when these are present. 

It is not unexpected that pronominal argument DPs may be null in 
Abaza, given the rich agreement system. Basic information is not lost 
when there is a null argument, since the relevant features of the null argu- 
ment are recoverable from agreement on the lexical head that licenses the 
null pronoun. This is consistent with the condition proposed in Rizzi 

(88) Let X be the licensing head of an occurrence of pro: then pro has the 

grammatical specification of the features on X co-indexed with it. 

Stative verbal morphology, Stative verbs generally express predicates 
which are states rather than actions or events. The classification of verbs as 
stative or dynamic is lexical, however, and does not conform exactly to the 
semantic distinction. A few verbs can occur with either stative or dynamic 
morphology with different interpretations based on this morphology. 

Stative verbs differ from dynamic verbs with respect to prefixes in that 
certain prefixes available for dynamic verbs are not available for stative 
verbs. These include directionals and other spatio-temporal prefixes. The 
stative system differs from the dynamic system with respect to suffixes in 
that stative verbs lack a number of distinctions available in the dynamic 
system. For example, stative morphology distinguishes only two tenses, 
past and present, in contrast to the eight seen in the dynamic system. Cer- 
tain moods are also unavailable in the stative system, for example the 

Most stative tense and mood suffixes are phonologically identical to 
those of dynamic tense and mood suffixes. Compare the dynamic yes-no 
question paradigm in (61a) and (61d) with its stative counterpart in (89). 

(89) a. Past y-J^-dma-Q-ma 

Did you (PL) have it? 

b. Present y-J^-9ma-w-ma 

Do you (PL) have it? 

32 Introduction and Background 

The most notable difference between dynamic and stative tense and 
mood morphology occurs with the indicative mood suffixes. The present 
stative indicative suffix is -b and the past stative indicative suffix is -n, as 
in (90). Tense and mood are combined in these suffixes. This contrasts 
with the dynamic indicative suffix -d, which requires a distinct tense suf- 
fix. (Compare examples (60a) and (60d) with (90a) and (90b), 

(90) a. Past y-s-dma-n 

I had it. 

b. Present y-$-9nia-b 

I have it. Cross-categorial morphology. Adverbs do not require or al- 
low agreement within the verbal complex (or elsewhere). There are two 
suffixes, which are probably better analyzed as clitics, that turn a phrase 
of any category into an adverb. These are la and ta. They are attached to 
the right of the last word in the phrase. The clitic la, exemplified in (91), is 
generally treated as an instrumental suffix in the literature. (My analysis 
in chapter 6 will treat Za as a type of postposition.) The clitic ta, exempli- 
fied in (92), has been treated in diverse ways in the literature, including 
as an adverbializer (Genko 1955), as a factitive case marker indicating 
what the host phrase becomes (Serdyuchenko 1956), and as a verbal suf- 
fix (Sergei Pazov, pers. comm.). In actual fact, the range of usage for both 
these clitics is much wider. (See example 23.) An account of them is be- 
yond the scope of this study. The crucial point for present purposes is that 
no agreement is required or allowed with phrases having these 

(91) ha-ft'ax^-la dBzda y-a^s 

1 p-after-AV who AWH-pass 

Who passed by behind us? 

(92) sahdy-tja dd-Maza-n 
baby-AV 3sr-3sf-raise-P§T 

She brought him up as a baby/since he was a baby. NART, 50 

SIL International and 

The University of Texas at Arlington 

Publications in Linguistics 

Publication 138 

Publications in Linguistics is a series published jointly by 
SIL International and the University of Texas at 
Arlington. The series is a venue for works covering a 
broad range of topics in linguistics, especially the 
analytical treatment of minority languages from all parts 
of the world. While most volumes are authored by 
members of the Institute, suitable works by others will 
also form part of the series. 

Series Editors 

Donald A. Burquest Mary Ruth Wise 

University of Texas at Arlington SIL International 

Volume Editors 

Mary Ruth Wise 
Rhonda Kartell 

Production Staff 

Bonnie Brown, Managing Editor 
Virginia Larson, Proofreader 

Margaret Gonzalez, Compositor 
Hazel Shorey, Graphic Artist